The long-term objective of this project is to understand how muscles convert chemical energy into mechanical force, motion, and work. All muscles contract via the motor-protein myosin, cyclically hydrolyzing ATP and pulling on actin filaments. The individual atomic structures of myosin and actin are known, but the structural changes myosin undergoes while it is pulling on actin are still poorly understood. Our work seeks to directly visualize, at the quasi-atomic level, the full range of structural changes myosin undergoes while working. The most interesting states, i.e. when myosin is generating force, are inherently dynamic but can be cryo-trapped and studied within the sarcomere, the specifically evolved lattice of thick and thin filaments that holds myosin and actin in close proximity and high density. Our approach remains unique worldwide and integrates muscle physiology, X-ray diffraction (XRD), rapid (~1 ms) cryo-preservation, electron microscopy, 3D reconstruction using electron tomography (EM/ET), class averaging, and quasi-atomic model building using the flight muscles from Lethocerus, whose nearly crystalline filament lattice make it the best possible specimen for such structural studies. Because individual myosins are not synchronized with each other, steady-state contraction automatically contains the full ensemble of all possible myosin states reflecting the biochemical equilibrium dynamics, open to biochemical or mechanical perturbation and investigation. Muscle physiology and XRD report the functional and structural status of the dynamic ensemble average, and are complemented by rapid cryo-trapping and EM/ET that allows us to directly visualize the entire ensemble in 3D. Multivariate data analysis identifies self-similar myosin structures (classes) that can be class-averaged, to increase resolution sufficiently for building unambiguous, crystal-derived atomic structures into the 3D densities and to quantify the relative populations of structural variants, which can then be related back to the biochemical equilibrium. This integrated approach is mutually cross-validating: physiology and XRD results are interpreted by structural models that are directly validated by EM/ET, whereas successful capture of the desired state is validated by monitoring force up to the moment of time-resolved cryo-trapping, and retention of native structure by EM/ET is validated by XRD monitoring of EM processing. Specific aims are to biochemically or mechanically perturb steady-state contractions to select among competing models of contraction by: 1) using EM/ET to directly image myosin structure; 2) using rapid mechanics and XRD to probe the ensemble average and cross-validate EM/ET results; 3) determining the structure of relaxed Lethocerus thick filaments; and 4) incorporating chimeric thin filaments into muscles to cross-correlate the mechanisms by which different muscles respond to stretch. A mechanistic and molecular understanding of contraction is a necessary prerequisite for comprehensive models of muscle function, both skeletal and cardiac, and for understanding how these mechanisms are deficient in human disease, including heart disease, muscle myopathies, muscle injuries, and sarcopenia.